A method for weakening site effect of brain function network based on wavelet transform

Abstract

The application discloses a brain function network site effect weakening method based on wavelet transform, relates to the technical field of biomedical signal processing, and unifies sampling frequency and electrode sites, carries out wavelet transform on converted electroencephalogram signals, decomposes the wavelet signals into wavelet signals of different frequencies, calculates average amplitudes of the wavelet signals of different frequencies, modulates the average amplitudes of the wavelet signals of different frequencies by using a ComBat algorithm, constitutes broadband electroencephalogram signals by using the modulated wavelet signals of different frequencies, extracts signal components of each rhythm frequency band in the broadband electroencephalogram signals, and constructs a brain function network by calculating the correlation between electrode sites. The application effectively weakens or eliminates the main site effect involved in the construction of the cross-site brain function network, and solves the site effect problems of inconsistent electrode site distribution, inconsistent amplitude-frequency response characteristics, collected samples and environmental variables and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biomedical signal processing, and in particular to a brain function network site effect weakening method based on wavelet transform. BACKGROUND

[0002] In multi-site brain function connectivity research projects, site effect refers to systematic bias introduced by differences in data collection locations or equipment. This effect can significantly affect the comparability of data and the accuracy of analysis. Site effects can be caused by many factors, including hardware equipment, data collection protocols, subject population characteristics, and environmental factors.

[0003] In brain function connectivity research based on electroencephalogram signals, different sites may use different types of electroencephalogram signal acquisition equipment, including different types of electroencephalogram caps, different models of electroencephalogram signal amplifiers, different numbers of electrodes, and electrode position distributions arranged according to different standards, ultimately resulting in differences in data quality and characteristics. The design parameters of different electroencephalogram signal amplifiers will result in different signal amplitude-frequency responses. The circuit design of high-pass and low-pass filters determines the sensitivity to different frequency signals, and the high-pass filter attenuates low-frequency drift, and the low-pass filter suppresses high-frequency noise. The configuration of the filter and the gain circuit, and the parameters of the digital-to-analog converter together determine the amplitude-frequency response of the collected signal. In addition, the research subjects at different sites may differ in demographic variables (such as age, gender, race, cultural background) or health status. The collection environment (such as noise level, temperature, humidity) and the training level of the operator may also introduce slight bias. These site effects can cause differences in functional connectivity patterns and mask the actual physiological or cognitive signals; reduce the effectiveness of statistical analysis, and the differences between sites can lead to false positive or false negative results; the incomparability of data between different sites can affect the feasibility of multi-center research and the universality of the results. SUMMARY

[0004] In order to overcome the defects in the prior art described above, the present application provides a brain function network site effect weakening method based on wavelet transform, which solves the problems of inconsistent electrode site distribution, inconsistent amplitude-frequency response characteristics, and site effects such as collection sample and environmental variables.

[0005] To achieve the above purpose, the technical scheme adopted by the present application comprises:

[0006] A brain function network site effect weakening method based on wavelet transform, comprising the following contents:

[0007] Select a standard sampling frequency to unify the electroencephalogram signals of each site to the standard sampling frequency; the electroencephalogram signals of the site are composed of signals from each electrode site collected by the site equipment;

[0008] selecting a target electrode site, converting electrode sites of each station to the target electrode site, thereby converting electroencephalogram signals of each station to obtain converted electroencephalogram signals;

[0009] For each station, the converted electroencephalogram signals of the station are wavelet transformed to be decomposed into wavelet signals of different frequencies; the wavelet signals of each frequency are subjected to ComBat modulation; the modulated wavelet signals of each frequency are composed to be broadband electroencephalogram signals; thereby obtaining broadband electroencephalogram signals of each station for constructing a brain function network.

[0010] Preferably, a finite impulse response filter is used to unify the electroencephalogram signals of each station to a standard sampling frequency.

[0011] Preferably, the lowest sampling frequency value in all station devices is selected as the standard sampling frequency, and the electroencephalogram signals with a sampling frequency higher than the standard sampling frequency are down-sampled to the standard sampling frequency.

[0012] Preferably, spherical interpolation is used for interpolation of electrode sites of the station to insert the target electrode site, thereby converting the electroencephalogram signals of each station to obtain the converted electroencephalogram signals, as shown below:

[0013] S T = M·S o

[0014]

[0015] wherein S T is the converted electroencephalogram signal of the station to be converted; S o is the original electroencephalogram signal of the station to be converted; M is a conversion matrix; elements m to in the conversion matrix represent the mapping relationship between the target electrode site t and the electrode site o in the station to be converted; θ to is the spherical angle between the target electrode site t and the electrode site o in the station to be converted; P n (·) is an n-order Legendre polynomial, and N is the order.

[0016] Preferably, the electrode site of one of the stations is selected as the target electrode site, the selected station is recorded as a target station, and the remaining stations are recorded as stations to be converted, and the electrode sites of the stations to be converted are recorded as electrode sites to be converted; the electrode sites to be converted are converted to the target electrode site, thereby converting the electroencephalogram signals of the stations to be converted to obtain converted electroencephalogram signals; as shown below:

[0017] The electrode sites to be converted are standardized to the spatial scale of the target station, and the calculation formula is as follows:

[0018]

[0019] wherein, the upper index o represents the coordinate system of the to-be-converted station, i.e., the to-be-converted coordinate system, the upper index t represents the coordinate system of the target station, i.e., the target coordinate system; x o , y o , z o are the coordinates of the to-be-converted electrode sites in the to-be-converted coordinate system; x t , y t , z t are the coordinates of the to-be-converted electrode sites in the target coordinate system;

[0020] For the to-be-converted station or the target station, the lower index c represents the center point of all electrodes in the station corresponding to each coordinate axis, i.e., the station center, is the target station center coordinate, is the to-be-converted station center coordinate, which is calculated as the average of the coordinates of all electrode sites in the station, N t is the number of electrodes of the target station, N o is the number of electrodes of the to-be-converted station;

[0021] The lower index + represents the electrode site with the largest positive distance from the station center corresponding to the coordinate axis, and the lower index - represents the electrode site with the largest negative distance from the station center corresponding to the coordinate axis, are the coordinates of the electrode sites with the largest positive distance from the target station center along the x, y, and z axes, respectively, are the coordinates of the electrode sites with the largest negative distance from the target station center along the x, y, and z axes, respectively, are the coordinates of the electrode sites with the largest positive distance from the target station center along the x, y, and z axes, respectively, are the coordinates of the electrode sites with the largest negative distance from the target station center along the x, y, and z axes, respectively.

[0022] The standardized to-be-converted electrode sites are interpolated using a spherical interpolation method and inserted into the target electrode sites, thereby converting the electroencephalogram signal of the to-be-converted station to obtain a converted electroencephalogram signal, which is specifically as follows:

[0023] S T = M·S o

[0024]

[0025] wherein, S T is the converted electroencephalogram signal of the to-be-converted station; S o is the original electroencephalogram signal of the to-be-converted station; M is a conversion matrix; the elements mto denotes the mapping relationship between the target electrode site t and the electrode site o in the to-be-transformed site; θ to is the spherical angle between the target electrode site t and the electrode site o in the to-be-transformed site; P n (·) is an n-order Legendre polynomial, and N is the order.

[0026] Preferably, for each site, the transformed electroencephalogram of the site is subjected to wavelet transform to be decomposed into wavelet signals of different frequencies, and after Hilbert transform, the average amplitude f of the wavelet signals of each frequency is obtained, and the calculation formula is:

[0027]

[0028] wherein, H(s) l is the value of the Hilbert transform of the wavelet signal s at the sampling point l, and L is the signal length;

[0029] At the same electrode site of each site, the average amplitude f of the wavelet signals of the same frequency of all subjects is obtained, the average amplitude f is subjected to ComBat modulation, and the modulated wavelet signals of each frequency are obtained, and the calculation formula is:

[0030]

[0031] wherein, is the modulated amplitude of the wavelet signal of frequency k of subject j at site i; f ijk is the amplitude of the wavelet signal of frequency k of subject j at site i; X ij is a covariate matrix carrying the physical information of the subject, and is the empirical Bayes estimation parameter of the regression model, and is the bias term coefficient;

[0032] Preferably, the following contents are further included: the modulated wavelet signals of each frequency are reconstructed into Z rhythm frequency bands, wherein the frequency range of the first rhythm frequency band K1 is [0, k1], the frequency range of the second rhythm frequency band K2 is (k1, k2],..., and the frequency range of the Zth rhythm frequency band K Z is (k Z-1 ,k Z ), 0 < k1 < k2 <... < k Z-1 < k Z ;

[0033]

[0034] wherein, for the reconstructed zth rhythm band signal, f k and are the average amplitudes of the wavelet signal of frequency k before and after modulation, respectively, c h,k is the component coefficient at level h and frequency k in wavelet packet decomposition, φ h,k (t) is the corresponding wavelet basis function, K z is the zth rhythm band, z = 1, 2,..., Z.

[0035] Preferably, for each rhythm band, the correlation corr ab between any two electrode sites a and b of each station is calculated for the reconstructed signals of the two electrode sites a and b.

[0036]

[0037] where s a , s b are the reconstructed signals of the two electrode sites a and b, are the mean values of s a , s b .

[0038] Based on the correlation between any two electrode sites a and b, the brain function network of each station at each rhythm band is obtained, wherein if the correlation corr ab between the two electrode sites a and b is greater than a set threshold, it indicates that the two electrode sites a and b are correlated, and a and b are connected as an edge.

[0039] The application further provides an electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the processor implements the above-mentioned wavelet transform-based brain function network station effect weakening method when executing the computer program.

[0040] The application further provides a computer program product comprising computer programs / instructions, which, when executed by a processor, implement the above-mentioned wavelet transform-based brain function network station effect weakening method.

[0041] The application has the following advantages:

[0042] (1) The application proposes a station effect weakening method, which effectively weakens or eliminates the main station effects involved in the construction of the cross-station brain function network through spherical interpolation, wavelet signal decomposition and reconstruction, ComBat signal modulation, and the like, and solves the problems of inconsistent electrode site distribution, inconsistent amplitude-frequency response characteristics, collected sample and environmental variables, and the like.

[0043] (2) The present application provides a systematic method flow that can effectively weaken or eliminate the main site effects involved in the construction of cross-site brain functional networks. These site effects include but are not limited to hardware device differences, data acquisition protocol differences, subject group characteristics differences, and environmental factor differences.

[0044] (3) By unifying the sampling rate, standardizing the electrode sites, and modulating the signal amplitude, the present application significantly improves the comparability of data between different sites, ensuring the reliability and universality of multi-center research results.

[0045] (4) Under the premise of ensuring signal quality, the present application uses advanced signal processing techniques such as wavelet transform and ComBat algorithm to ensure the consistency of the amplitude-frequency response of electroencephalogram signals and reduce signal distortion caused by device and environmental differences.

[0046] (5) By standardizing the research objects and collection environment of different sites, the present application effectively solves the differences in brain functional connectivity caused by demographic variables (such as age, gender, and cultural background) and environmental variables (such as noise level and temperature).

[0047] (6) The method of the present application can reduce the occurrence of false positive and false negative results, improve the effectiveness and accuracy of statistical analysis, and thus more accurately reflect the actual physiological or cognitive signals.

[0048] (7) By providing a unified signal processing method, the present application improves the feasibility of multi-center research, promotes the integration and analysis of cross-site data, and enhances the universality and application value of research results. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The flowchart of the method of the present application.

[0050] Figure 2 The schematic diagram of the electrode site distribution of the electroencephalogram cap.

[0051] Figure 3 The schematic diagram of the amplitude-frequency response comparison before and after modulation.

[0052] Figure 4 The schematic diagram of the brain functional network site effect comparison of each rhythm band before and after modulation. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0054] As shown in Figure 1 The embodiment provides a brain function network site effect weakening method based on wavelet transform, and comprises the following steps:

[0055] S1, uniform sampling frequency: selecting a standard sampling frequency, and unifying the electroencephalogram signals of the sites to the standard sampling frequency; the electroencephalogram signals of the sites are composed of signals of each electrode site collected by site equipment.

[0056] In the embodiment, the minimum value of the sampling frequencies of all site equipment is selected as the standard sampling frequency, and the electroencephalogram signals with a sampling frequency higher than the standard sampling frequency are down-sampled to the standard sampling frequency. A finite impulse response filter is used to unify the electroencephalogram signals of the sites to the standard sampling frequency.

[0057] The electroencephalogram signals of the sites are composed of signals of each electrode site collected by site equipment (an electroencephalogram cap);

[0058] S2, electrode site unification: selecting a target electrode site, converting the electrode sites of the sites to the target electrode site, thereby converting the electroencephalogram signals of the sites, obtaining converted electroencephalogram signals, and unifying the data collected by electroencephalogram caps with inconsistent electrode sites into data of the same electrode site.

[0059] S3, wavelet transform decomposition: performing wavelet transform on the converted electroencephalogram signals of the sites, decomposing the converted electroencephalogram signals into wavelet signals of different frequencies, and calculating average amplitudes of the wavelet signals of different frequencies.

[0060] S4, ComBat modulation: using a ComBat algorithm to modulate the average amplitudes of the wavelet signals of different frequencies.

[0061] S5, wide-spectrum electroencephalogram signal reconstruction: constructing a wide-spectrum electroencephalogram signal from the modulated wavelet signals of different frequencies.

[0062] S6, brain function network construction: extracting signal components of each rhythm frequency band in the wide-spectrum electroencephalogram signal, and constructing a brain function network by calculating the correlation between the signal components of each rhythm frequency band between electrode sites.

[0063] S7, brain function network edge modulation: using a ComBat algorithm to modulate all edges in the brain function network.

[0064] The embodiment uses the brain electrical data of 20-30 years old closed eyes resting state from three sites, wherein: site 1 contains 147 people, the male to female ratio is 15:34, the average age is 25.5 years old, the device sampling frequency is 1000 Hz, and the number of effective electrodes is 64; site 2 contains 97 people, the male to female ratio is 49:48, the average age is 22.7 years old, the device sampling frequency is 1000 Hz, and the number of effective electrodes is 62; site 3 contains 132 people, the male to female ratio is 7:5, the average age is 23.4 years old, the device sampling frequency is 512 Hz, and the number of effective electrodes is 64. Three sites use three different types of brain electrical amplifier devices. The electrode site distribution is as shown in Figure 2 The signal length is uniformly 30 s.

[0065] In the embodiment, the standard sampling frequency is selected as 512 Hz, and the brain electrical signals of all subjects in site 1 and site 2 are down-sampled to 512 Hz. The filter involved is a finite impulse response low-pass filter, the filter length is 101 (the order is 100), the cutoff frequency is 256 Hz, and the Nyquist frequency is 500 Hz.

[0066] In the embodiment, the electrode sites of site 1 are selected as the target electrode sites, site 1 is the target site, the electrode sites of site 1 are the target electrode sites, and site 2 and site 3 are the to-be-converted sites. The electrode sites of site 2 and site 3, i.e., the converted electrode sites, are standardized to the spatial scale of site 1, and the calculation formula is:

[0067]

[0068]

[0069] wherein, the upper subscript o represents the coordinate system of the to-be-converted site, i.e., the to-be-converted coordinate system, the upper subscript t represents the coordinate system of the target site, i.e., the target coordinate system; x o , y o , z o are the coordinates of the to-be-converted electrode site in the to-be-converted coordinate system; x t , y t , z t are the coordinates of the to-be-converted electrode site in the target coordinate system;

[0070] For the to-be-converted site or the target site, the lower subscript c represents the center point corresponding to each coordinate axis of all electrodes in the site, i.e., the site center, is the target site center coordinate, is the to-be-converted site center coordinate, and the calculation method is the mean value of all electrode site coordinates, N t is the number of electrodes of the target site, N o is the number of electrodes of the to-be-converted site;

[0071] The subscript + represents the electrode site with the largest distance from the site center along the positive direction of the corresponding coordinate axis, and the subscript - represents the electrode site with the largest distance from the site center along the negative direction of the corresponding coordinate axis. are respectively the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, and z axes in the positive direction, are respectively the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, and z axes in the negative direction, are respectively the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, and z axes in the positive direction, are respectively the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, and z axes in the negative direction.

[0072] The normalized electrode sites to be converted are interpolated using a spherical interpolation method and inserted into the target electrode sites, thereby converting the electroencephalogram signal of the site to be converted to obtain a converted electroencephalogram signal, which is specifically as follows:

[0073] S T = M·S o

[0074]

[0075] wherein S T is the converted electroencephalogram signal of the site to be converted; S o is the original electroencephalogram signal of the site to be converted; M is a conversion matrix; the element m to in the conversion matrix represents the mapping relationship between the target electrode site t and the electrode site o in the site to be converted; θ to is the spherical angle between the target electrode site t and the electrode site o in the site to be converted; P n (·) is an n-order Legendre polynomial, and N is the order, N = 50.

[0076] The signal of each electrode site of each subject is subjected to a discrete wavelet packet transform (WPT). In this embodiment, the target signal is decomposed into 50 wavelet signals by using a Daubechies 4 (db4) wavelet for a 1-50 Hz frequency band, and the frequency interval between each wavelet signal is 1 Hz.

[0077] After Hilbert transform, the average amplitude f of each frequency wavelet signal is obtained, and the calculation formula is:

[0078]

[0079] Wherein, H(s) l Let L be the value at sampling point l after performing a Hilbert transform on the wavelet signal s, where L is the signal length;

[0080] At the same electrode sites in each station, the average amplitude f of wavelet signals of the same frequency for all subjects was obtained. The average amplitude f was then modulated using ComBat to obtain the modulated wavelet signals for each frequency. The calculation formula is as follows:

[0081]

[0082] in, f is the modulated amplitude of the wavelet signal at frequency k of subject j at station i; ijk X represents the amplitude of the wavelet signal at frequency k of subject j at station i; ij It is a covariate matrix that carries the subject's vital signs information. and These are the empirical Bayesian estimates of the regression model. and It is the coefficient of the deviation term;

[0083] After modulation, the signal from each subject at each station is reconstructed into five rhythmic frequency bands: the first rhythmic frequency band K1 has a frequency range of [0, 4Hz], the second rhythmic frequency band K2 has a frequency range of (4Hz, 8Hz], the third rhythmic frequency band K3 has a frequency range of (8Hz, 13Hz], the fourth rhythmic frequency band K4 has a frequency range of (13Hz, 30Hz], and the fifth rhythmic frequency band K5 has a frequency range of (30Hz, 50Hz). The calculation formula is as follows:

[0084]

[0085] in, For the reconstructed z-th rhythm frequency band signal, f k and c represents the average amplitude of the wavelet signal at frequency k before and after modulation. h,k φ represents the component coefficients at level h and frequency k during wavelet packet decomposition. h,k (t) is the corresponding wavelet basis function, K z Let z be the z-th rhythm frequency band, where z = 1, 2, ..., 5.

[0086] like Figure 3 As shown, before reconstruction (refer to...) Figure 3 (See the top image). The signal amplitude-frequency response of station 2 is significantly different from the other two stations. After reconstruction (refer to...) Figure 3As shown in the following figure (see Fig. 2), the signal amplitude-frequency response difference of the three sites is significantly weakened, showing high consistency. The results show that this method effectively weakens the site effect existing in the signal amplitude-frequency response.

[0087] For each rhythm band, the correlation corr between the reconstructed signals of any two electrode sites a and b of each subject of each site is calculated ab , and the calculation formula is:

[0088]

[0089] wherein s a and s b are the reconstructed signals of the two electrode sites a and b, and s a and s b are the mean values of s

[0090] Based on the correlation between any two electrode sites a and b, the brain function network of a specific rhythm frequency based on correlation is obtained. If the correlation corr between the two electrode sites a and b is greater than a set threshold, it indicates that the two electrode sites a and b are correlated, and a and b are connected into an edge. ab

[0091] The signal correlation between each pair of electrode sites of all subjects of the three sites is normalized and corrected, and the statistical correction method adopted for the signal amplitude-frequency characteristics is the same. Here, the modulation method of ComBat (using the age and gender of the subject as the covariate) is also used, and the calculation formula is:

[0092]

[0093] wherein, is the modulated vth element of the upper triangular matrix of the connection matrix of site i and subject j, is the vth element of the upper triangular matrix of the connection matrix of site i and subject j, X ij is the covariate matrix, and are the empirical Bayes estimation parameters of the regression model, and are the coefficients of the bias terms to be eliminated.

[0094] As shown in the following figure (see Fig. 2), the signal amplitude-frequency response difference of the three sites is significantly weakened, showing high consistency. The results show that this method effectively weakens the site effect existing in the signal amplitude-frequency response. Figure 4 ​As shown, 10 heatmaps represent the comparison of results in five frequency bands before and after modulation. Each heatmap contains 64 rows and 64 columns, with each row and column corresponding to a signal channel. Each value is the p-value of the Kruskal-Wallis test result regarding the distribution of the correlation between the two electrode signals represented by the row and column among the subjects. Note that the values ​​on the diagonal of the heatmap are based on the correlation between each signal channel and itself; this correlation coefficient is always equal to 1 and therefore has no statistical significance. Figure 4 The value was set to 0. The null hypothesis of this test is that the correlations from the three sites are identically distributed among their respective subjects, meaning that when p < 0.05, the input distributions are different. Figure 4 Before ComBat modulation, numerous correlations with varying distributions were observed across all frequency bands. However, after modulation, the correlation distribution among any electrode sites in all subjects across all five rhythmic frequency bands showed no significant difference across the three sites. This result indicates that this method effectively weakens the site effect present in the construction of brain functional networks.

[0095] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for weakening the site effect of brain function network based on wavelet transform, characterized in that, The method comprises the following steps: Selecting a standard sampling frequency, and unifying the electroencephalogram signals of each site to the standard sampling frequency; the electroencephalogram signals of the site are composed of signals of each electrode site collected by a site device; Selecting a target electrode site, and converting the electrode sites of each site to the target electrode site, so as to convert the electroencephalogram signals of each site and obtain converted electroencephalogram signals; Using a spherical interpolation method to interpolate the electrode sites of the site and insert the target electrode site, so as to convert the electroencephalogram signals of each site and obtain converted electroencephalogram signals; Selecting an electrode site of one of the sites as a target electrode site, and recording the selected site as a target site and the remaining sites as to-be-converted sites; the electrode sites of the to-be-converted sites are recorded as to-be-converted electrode sites; Converting the to-be-converted electrode sites to the target electrode site, so as to convert the electroencephalogram signals of the to-be-converted sites and obtain converted electroencephalogram signals; For each site, performing wavelet transform on the converted electroencephalogram signals of the site to decompose the converted electroencephalogram signals into wavelet signals of different frequencies; Performing ComBat modulation on the wavelet signals of each frequency; Constructing the wavelet signals of each frequency after modulation into broadband electroencephalogram signals; and obtaining the broadband electroencephalogram signals of each site for constructing a brain function network; For each site, performing wavelet transform on the converted electroencephalogram signals of the site to decompose the converted electroencephalogram signals into wavelet signals of different frequencies, and obtaining average amplitudes f of the wavelet signals of each frequency after Hilbert transform. At the same electrode sites of each site, obtaining the average amplitudes f of the wavelet signals of the same frequency of all subjects, and performing ComBat modulation on the average amplitudes f to obtain the wavelet signals of each frequency after modulation.

2. The method of claim 1, wherein the method is based on wavelet transform. Using a finite impulse response filter to unify the electroencephalogram signals of each site to a standard sampling frequency.

3. The method for weakening the site effect of brain function network based on wavelet transform according to claim 1 or 2, characterized in that, Selecting the minimum value of the sampling frequencies of all site devices as the standard sampling frequency, and reducing the sampling frequencies of the electroencephalogram signals higher than the standard sampling frequency to the standard sampling frequency.

4. The method of claim 1, wherein the method is based on wavelet transform. Using a spherical interpolation method to interpolate the electrode sites of the site and insert the target electrode site, so as to convert the electroencephalogram signals of each site and obtain converted electroencephalogram signals; S T = M · S o Wherein, S T is the converted electroencephalogram signal of the to-be-converted electrode site; S o is the original electroencephalogram signal of the to-be-converted electrode site; M is a conversion matrix; the elements m to in the conversion matrix represent the mapping relationship between the target electrode site t and the electrode site o in the to-be-converted electrode site; θ to is the spherical angle between the target electrode site t and the electrode site o in the to-be-converted electrode site; P n (·) is an n-order Legendre polynomial, and N is the order.

5. The method of claim 1, wherein the method is based on wavelet transform. Selecting an electrode site of one of the sites as a target electrode site, and recording the selected site as a target site and the remaining sites as to-be-converted sites; the electrode sites of the to-be-converted sites are recorded as to-be-converted electrode sites; Converting the to-be-converted electrode sites to the target electrode site, so as to convert the electroencephalogram signals of the to-be-converted sites and obtain converted electroencephalogram signals; wherein the upper index o represents the coordinate system of the site to be converted, i.e. the coordinate system to be converted, and the upper index t represents the coordinate system of the target site, i.e. the target coordinate system; x o , y o , z o are the coordinates of the electrode site to be converted in the coordinate system to be converted; x t , y t , z t are the coordinates of the electrode site to be converted in the target coordinate system; For the site to be converted or target site, subscript c represents the center point of the site corresponding to each coordinate axis, i.e. the site center, is the target site center coordinate, is the site to be converted center coordinate, calculated as the mean of the site's electrode site coordinates, N t is the number of electrodes of the target site, N o is the number of electrodes of the site to be converted; subscript + represents the electrode site with the largest distance from the site center along the positive direction of the corresponding coordinate axis, and subscript - represents the electrode site with the largest distance from the site center along the negative direction of the corresponding coordinate axis, respectively are the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, z axis positive direction, respectively are the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, z axis negative direction, respectively are the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, z axis positive direction, respectively are the coordinates of the electrode sites in the target site with the largest projection distance from the target site center along the x, y, z axis negative direction, For each site, performing wavelet transform on the converted electroencephalogram signals of the site to decompose the converted electroencephalogram signals into wavelet signals of different frequencies, and obtaining average amplitudes f of the wavelet signals of each frequency after Hilbert transform. S T = M · S o Wherein, S T is the converted electroencephalogram signal of the to-be-converted electrode site; S o is the original electroencephalogram signal of the to-be-converted electrode site; M is a conversion matrix; the elements m to in the conversion matrix represent the mapping relationship between the target electrode site t and the electrode site o in the to-be-converted electrode site; θ to is the spherical angle between the target electrode site t and the electrode site o in the to-be-converted electrode site; P n (·) is an n-order Legendre polynomial, and N is the order.

6. The method of claim 1, wherein the method is based on wavelet transform. At the same electrode sites of each site, obtaining the average amplitudes f of the wavelet signals of the same frequency of all subjects, and performing ComBat modulation on the average amplitudes f to obtain the wavelet signals of each frequency after modulation. where H(s) = 1 - s l is the value of the Hilbert transform of the wavelet signal s at sample point l, and L is the signal length. On the same electrode site of each site, the average amplitude f of the wavelet signal of the same frequency of all subjects is obtained, the average amplitude f is subjected to ComBat modulation, and the modulated wavelet signal of each frequency is obtained, and the calculation formula is: where, is the modulated amplitude of the wavelet signal of frequency k at site i for subject j; f ijk is the amplitude of the wavelet signal of frequency k at site i for subject j; X ij is the covariate matrix, carrying the subject's physical information, and is the empirical Bayes estimate parameter of the regression model, and is the bias term coefficient.

7. The method of claim 1 or 6, wherein the method is based on wavelet transform. Further comprising the following: wavelet signal reconstruction of each frequency after modulation into Z rhythm frequency bands, wherein the frequency range of the first rhythm frequency band K1 is [0, k1], the frequency range of the second rhythm frequency band K2 is (k1, k2],..., the frequency range of the Zth rhythm frequency band K Z is (k Z-1 , k Z ), 0 < k1 < k2 <... < k Z-1 < k Z ; wherein is the reconstructed z-th rhythmic band signal, f k and are the pre- and post-modulation average amplitude of the wavelet signal at frequency k, c h,k is the component coefficient at level h and frequency k during wavelet packet decomposition, φ h,k (t) is the corresponding wavelet basis function, K z is the z-th rhythmic band, z = 1, 2,..., Z.

8. The method of claim 7, wherein the method is based on wavelet transform. For each rhythm band, the correlation corr between any two electrode sites a and b of each station is calculated for the reconstructed signals of the two electrode sites a and b ab The calculation formula is: where s a , s b are the reconstructed signals of the two electrode sites a and b, respectively, are the mean values of s a , s b , respectively. Based on the correlation between any two electrode sites a and b, the brain function network of each site in each rhythm frequency band is obtained, wherein if the correlation corr ab between the two electrode sites a and b is greater than a set threshold, it indicates that the two electrode sites a and b are correlated, and a and b are connected into an edge.

9. An electronic device, comprising: The computer program is stored in the memory and can be run on the processor, and the processor executes the computer program to realize the wavelet transform-based brain function network site effect weakening method in any one of claims 1-8.

10. A computer program product, characterised in that, The computer program / instructions are executed by the processor to realize the wavelet transform-based brain function network site effect weakening method in any one of claims 1-8.

Images